The New Palgrave Dictionary of Economics

Living Edition
| Editors: Palgrave Macmillan

Economic Demography

  • Allen C. Kelley
  • Robert M. Schmidt
Living reference work entry


Economic demography is an area of study that examines the determinants and consequences of demographic change, including fertility, mortality, marriage, divorce, location (urbanisation, migration, density), age, gender, ethnicity, population size and population growth. This article reviews and critically evaluates important macroeconomic dimensions of the ‘population debates’ between the ‘optimists’ and the ‘pessimists’ since 1950. It concludes with an examination of demography in the popular ‘convergence’ growth models of the 1990s.


Adult equivalency Ageing Agricultural growth and population change Capital accumulation Convergence Demographic drag Demographic gift Demographic transition Diffusion of technology Diminishing returns Dismal science Economic demography Economic development Economic growth Education Endogenous growth Fertility Free rider problem Human capital Innovation Kuznets, S. Labour productivity Learning-by-doing Life expectancy Life-cycle modelling Malthus., T. R. Mortality Population density Population growth Population size Renewable resources Research and development Rule of law Saving Simon, J. L. Subsistence Technical change 

JEL Classifications


Economic demography is an area of study that examines the determinants and consequences of demographic change, including fertility, mortality, marriage, divorce, location (urbanisation, migration, density), age, gender, ethnicity, population size, and population growth. An applied area of research, economic demography draws upon the theoretical and applied fields of economics. For example, the determinants of fertility or migration primarily draw upon microeconomic theory and labour economics, while the consequences of population growth or ageing primarily draw upon macroeconomic theory and development economics.

The field has had a long tradition of controversy, beginning with the publication in 1798 of An Essay on the Principle of Population by the Reverend Thomas Malthus. The basic Malthusian model is founded on two propositions: (a) population, when unchecked, increases at a geometric rate (for example, 1, 2, 4, 8…) and (b) food, in contrast, expands at an arithmetic rate (for example, 1, 2, 3, 4…). The result is a population trapped at a meagre standard of living. Short of ‘preventive checks’ (birth control), population is constrained to live at subsistence by ‘positive checks’ (deaths, war, famines and pestilence). In later writings Malthus admitted the possibility of ‘moral restraint’ that could deter births, primarily through the postponement of marriage. However, he held little hope for a notable attenuation of the ‘natural passions’ of the working class.

While much of the controversy relating to Malthusianism has focused on the determinants of population growth, a second premise of his model relates to its economic underpinnings: the determinants of agricultural growth. Here Malthus appealed to the historical law of diminishing returns in agriculture. While this proposition engendered relatively little dispute at the time, history has since documented widespread and sometimes notable improvements in agricultural technology. Indeed, food production has represented an engine of growth in many of the areas that Malthus investigated. In some areas today, governments worry about ‘excess’ food production that depresses prices and farmers’ living standards. Unfortunately, the pessimistic food-production predictions, when confronted by rapid population growth, caused economics to be dubbed the ‘dismal science’.

The enormous popularity of the Malthusian ideas was the result of several factors: the model’s simplicity and its explanation of poverty (the poor failed to exercise moral restraint, ending up with large families); the appeal of the message that subsidising the poor is of questionable efficacy; and the plausibility of the Malthusian argument given the unexpected ‘population explosion’ revealed by the 1801 census. These and other elements of the ‘Malthusian debate’ provide a useful taxonomy for organising the present article.

Specifically, we highlight the macroeconomic dimensions of the economic consequences of population growth since 1950. As with the early Malthusian debates, an assessment of the macroeconomic impacts of demographic change on economic production has resulted in an outpouring of research, which has spawned further debate. There are periods when vigorous Malthusian-like alarmism has carried the day; there are periods of counter-challenges; and, since the mid-1980s, there has been a productive ‘revisionist’ movement. In short, the simplistic Malthusian notion of diminishing returns in production has given way to more informed modelling of economic–demographic interactions. An assessment of the historical evolution of this literature will constitute the bulk of this review and appropriately delimits the scope of our essay since a wide range of important microeconomic themes are taken up in other articles in this dictionary (see “ Fertility in Developing Countries,” “ Family Decision Making,” “ Marriage and Divorce,” and “ Retirement,” and multiple articles dealing with the topics of gender, ageing and mortality).

We begin by examining population impacts in one-sector growth models. This leads nicely into a more detailed assessment of factor accumulation, and in particular, the impacts of demography on saving, investment and technological change. This is in turn followed by an analytical description of the evolution of economic–demographic thinking since 1950. Such a perspective exposes many of the key analytical and empirical linkages of interest. The article concludes with an examination of ‘convergence modelling’, a useful paradigm that exposes the roles of changing demographic structures that take place over the demographic transition.

Theory: Modelling Economic–Demographic Change

One-Sector Growth Models

The aggregate production function constitutes the primary organising device for delineating the impacts of demographic change on economic growth. Within this model, labour productivity depends on the availability of complementary factors of production (land, natural resources, human and physical capital) and technology. If we assume, for convenience, that labour is a constant fraction of population, then population size directly affects aggregate output.

In a production function with constant returns to scale, an increase in population growth will lower the average availability of other factors of production – a ‘resource-shallowing’ effect, and, through diminishing returns, reduce the growth of worker productivity. Such an adverse demographic impact can be magnified (or attenuated) if population growth diminishes (raises) the growth rate of complementary factors.

In a standard growth model with factor inputs of labour and capital, and a saving rate and pace of technological change that are exogenous with respect to population growth, demography affects the long-run level but not the long-run growth rate of output per capita. This is because the capital-shallowing effect of increased population will eventually reduce the capital per worker ratio to a level sufficient to be maintained by a fixed rate of saving. In this case, long-run growth is determined by the pace of technological change. The determinants of the ‘fixed’ saving rate and pace of technology growth, both considered in more detail below, are central to the analysis.

If one relaxes some of the assumptions of this model, the impact of population growth on per capita output growth can be ambiguous. Negative impacts can arise through diminishing returns, diseconomies of scale, and perhaps savings, while positive impacts can arise through induced technological change, economies of scale, and possibly savings. Most economists believe that adverse capital-shallowing impacts will dominate positive feedback effects, although the magnitude of the demographic impacts may not be all that large.


Possibly the most investigated linkage of population growth to economic growth has been the impact of demographic change on saving. Two perspectives dominate.

Adult equivalency. Rapid (slow) rates of population growth result in a disproportionate number of children (elderly adults) who consume, but contribute relatively little to, household income. In recognising that these ‘dependents’ consume less than a working-age adult, the notion of an ‘adult equivalent’ consumer was born. The financing of an additional child’s ‘adult-equivalent’ consumption has been hypothesised to be out of saving. Such a view, however, has been challenged by consideration of several offsetting alternatives. Specifically, children may (a) substitute for other forms of consumption, (b) contribute directly to household market and non-market income, (c) encourage parents to work more (or less), (d) stimulate the amassing (or reduction) of estates, and (e) encourage (or discourage) the accumulation of certain types of assets (for example, education or farm implements). The net impact of changing dependency rates on saving is therefore theoretically ambiguous. This is particularly the case if one views human capital as an investment financed in part by households and governments. At any rate, empirical evidence showing negative impacts of youth dependency on saving are found in several studies.

The life-cycle. A second population-saving linkage is based on a life-cycle formulation incorporated into a lifetime household utility function. Specifically, households attempt to even out their lifetime consumption by setting aside earnings during working years to finance consumption by their children as well as for their own retirement. This formulation can yield positive or negative impacts on aggregate saving depending on the relative sizes of the dissaving youth and elderly cohorts. While empirical evidence from life-cycle modelling is mixed, those studies do tend to show linkages between age structure and saving. However, the direction and magnitude of that impact depends upon time and place. (See, for example, Mason 1987; Higgins 1998; and Lee et al. 2001.)

Population-Sensitive Government Spending

Government spending on population-sensitive activities such as schooling (youth) and health (elderly) has been alleged both to reduce saving and to crowd out spending on relatively growth-oriented investments. These two hypotheses constitute the core of Ansley J. Coale and Edgar M. Hoover’s (1958) path-breaking study of India. While these premises are appealing, they require qualification. Governments have many options to accommodate population pressures. Indeed, limited empirical evidence (for example, Schultz 1987) has shown that education financing can be met all or in part by (a) trade-offs within the public sector, (b) reductions in per pupil expenditures, and (c) efficiency gains. While the second approach can be expected to reduce the quality of education (and therefore future productivity), the importance of population pressures on government spending or educational quality is uncertain.

Technological Change: Density, Size and Endogenous Growth

While development economists have for decades harkened the pace of technological change as a (the?) major source of economic growth, most standard growth theory models take the rate of technological change as exogenous. With technological change independent of demographic change, population growth per se will have no impact on the pace of economic growth in long-run equilibrium. By contrast, if technological change is all or in part embodied in new investment, then a vintage specification is appropriate whereby new capital is relatively more productive than old. In this set-up, population growth can be economic-growth enhancing by expanding the rate at which technology is incorporated into production. In yet another specification, population growth can directly affect the rate of technological change and/or its form (factor bias). Kenneth J. Arrow (1962) has hypothesised that learning by doing is quickened in an environment of rapid employment growth.

A fourth linkage between technology and demography is found in ‘endogenous growth’ models that relate the pace of technology directly to population size. In particular, the benefits of R&D are assumed to be available to all firms without cost; that is, an R&D industry generates a non-rival stock of knowledge. As a result, if we hold constant the share of resources used for research, an increase in population size advances technological change without limit. This somewhat controversial prediction has been qualified by models that incorporate various firm- or industry-specific constraints on R&D production. Such models typically reduce, but do not eliminate, the positive impacts of population size which, as in the embodiment models above, are manifested largely during the ‘transition’ to long-run equilibrium.

Evidence on the roles of demographic-technology linkages and growth has been fragmentary and sparse. A pioneering study by Hollis Chenery and Moises Syrquin (1975) draws upon the experience of 101 countries across the income spectrum over the period 1950–70. They find that the structure of development reveals strong and pervasive scale effects (measured by population size) that vary by stage of development. Basically, small countries develop a modern productive industrial structure more slowly and later, while large countries have higher levels of accumulation and (presumably) higher rates of technological change. Although these roles for demography may have been important historically, the impacts plausibly have waned somewhat: (a) economies in infrastructure are judged to be substantially exhausted in cities of moderate size; (b) specialisation through international trade provides a means of garnering some or many of the benefits of size; and (c) scale effects are most prevalent in industries with relatively high capital–labour ratios and such industries are inappropriate to the factor proportions of developing countries.

It is in agriculture where the positive benefits of population size have been most discussed. Higher population densities can lower per unit costs and increase the efficiency of transport, irrigation, extension services, markets and communications (Glover and Simon 1975). Possibly the most cited work is that by Ester Boserup (1965, 1981), who observes that increasingly productive agricultural technologies are made economically attractive in response to higher land densities. While this is probably true, the issue becomes one of identifying the quantitative magnitude of such effects over varying population sizes and in differing institutional settings. One must be cautious in attributing causation. For example, while high population densities may have accounted for a portion of expanded agricultural output in recent decades, in several important Asian countries these densities were sufficiently high decades ago to justify the investments associated with the new technologies. Boserup in more recent writing has been less sanguine about the benefits of population size because densities appropriate to modern technologies in Asia are three to four times the average for Africa and Latin America.

In short, a wide-ranging review of the literature does not provide a strong consensus on the quantitative linkages between the size and growth of population, on the one hand, and the pace of technological change and economic growth, on the other hand.

The Bottom Line

An evaluation of population growth on economic growth through the filter of formal economic-growth modelling yields limited results: population growth affects the level but not the growth of per capita income in long-run equilibrium. Moreover, the key determinants of long-run growth are saving and technology. Only if these factors depend on demographic change does population matter. This somewhat constraining limitation of growth theory has caused researchers to branch out and explore a host of economic–demographic interactions using less formal paradigms. This blossoming literature has been extensive, lively and sometimes contentious.

Evolution of Population-Impacts Thinking: 1950–90

Four major studies, two by the United Nations (1953, 1973) and two by the National Academy of Sciences (1971, 1986), reveal well the evolution of thinking on population matters over the period 1950–90. Three individual scholars, Coale and Hoover and Simon, also played prominent and important roles. (This section draws on Kelley 2001.)

United Nations, 1953

The 1953 United Nations report, Determinants and Consequences of Population Trends, easily represents the most important contribution to population thinking since the writings of Malthus. Unlike Malthus, however, the UN study was balanced and exhaustive both in detail and in coverage. Some 21 linkages between population and the economy were taken up. For example, the impacts of population on the economy can be: (a) positive due to economies of scale and organisation; (b) negative due to diminishing returns; or (c) neutral due to technology and social progress. An evaluation of these and other linkages led to a mildly negative overall assessment that was both cautious and qualified.

The most notable feature of this report was its methodology. More than any major study on population to that time, the UN Report embraced a methodology that would ultimately represent elements of modern-day ‘revisionism’. Specifically, the report (a) downgraded the importance of population growth’s impact on economic growth by placing it on a par with several other determinants of equal or greater impact; (b) assessed the consequences of population over a long period of time; and (c) emphasised the importance of feedbacks within and between the economic and political systems.

Coale and Hoover, 1958

The next major contribution to the population-impacts literature was provided by Ansley J. Coale and Edgar M. Hoover in their 1958 book Population Growth and Economic Development in Low-Income Countries. Based on simulations of a mathematical model calibrated with Indian data, they concluded that India’s development would be enhanced by lower population growth. This was due to the hypothesised adverse impacts of population on household saving. It was also proffered that ‘unproductive’ investments in human capital (such as health and education) would partially displace investments in ‘relatively productive’ forms (such as machines and factories). Economic growth would diminish in response.

Empirically, the above hypotheses have not been convincingly established. While several studies have exposed negative dependency-rate impacts on saving, there are others that show little or no impact. Overall, the findings are mixed, with a tilt toward supporting the Coale and Hoover formulation. (See section “Saving” above for a discussion of the trade-offs that households can make to maintain saving in response to expanding family size.)

Similarly, there are alternative ways for governments to organise and finance schooling in response to population pressures. Unfortunately, studies of this are limited, although one by T. Paul Schultz (1987) finds no support for the Coale and Hoover (1958) formulation.

National Academy of Sciences, 1971

Arguably the most pessimistic assessment of the consequences of population growth was a study compiled by the National Academy of Sciences (NAS). The panel’s final submission, Rapid Population Growth: Consequences and Policy Implications, issued in 1971, appeared in two volumes: volume 1, Summary and Recommendations, and volume 2, Research Papers. Unfortunately, the Summary volume appeared to be more political than academic in goal and orientation, and was not faithful to many of the underlying research reports assembled by the panel. Indeed, the Summary volume highlighted some 25 alleged negative consequences of population growth, whereas it downplayed or eliminated impacts that could be considered as ‘neutral’ or ‘favourable’. As a result, the Summary represents an upper bound on the negative consequences of population growth. (A detailed documentation exposing the somewhat controversial way in which the Summary was compiled is provided by Kelley 2001.)

What can be learned from the NAS study? First, given its apparent bias and the lack of a systematic vetting of volume 1 by members of the panel, it is difficult to use that volume, either in full or in part. However, the individual papers are available and they, in total, offer a more balanced treatment. Second, by its own acknowledgment, the study focused on the short run when negative impacts of population change are most likely to prevail. (‘We have limited ourselves to relatively short term issues’; 1971, p. vi.) By contrast, ‘direct’ (short-run) impacts of demographic change are almost always attenuated (and sometimes offset) by ‘indirect feedbacks’ that occur over longer periods of time. Thus the decision by the NAS panel to focus only on the short-run direct impacts resulted in an overly negative assessment of the consequences of population growth.

Third, economists were underrepresented on both the panel and in providing background reports. This is relevant since economists have substantial faith in the capacity of markets, individuals and institutions to adjust in the face of population pressures. Such adjustments, of course, take time and they are not without cost. Finally, this NAS Report provides a striking example of the difficulty of maintaining objectivity when social science research enters the public policy domain.

United Nations, 1973

In 1973 the United Nations weighed in with an update of its previous seminal work (United Nations 1953). In contrast to the broadly eclectic stance in the earlier report, the new one ended with a mild to moderate negative overall assessment of rapid population growth. The authors were concerned with the ability of agriculture to feed expanding populations (à la Malthus) and the difficulty of offsetting capital shallowing (à la Coale and Hoover). Still, the 1973 Report, whose conclusions are highly qualified, is not alarmist, nor is it all that pessimistic. The reason for this moderate stance was the exceptionally influential empirical finding of Simon Kuznets (1960, pp. 19–20, 63) that notable negative correlations between population growth and per capita output growth were largely absent in the data. Given the strong priors of some contributors to the UN study, a failure to find a negative association in the aggregate data by a scholar with impeccable credentials had a profound impact. Indeed, this singular finding arguably kept the population debate alive for yet another round of assessments in the 1980s.

Revisionism, 1980s and Beyond

The 1980s represented a decade when many of the underlying assumptions and conclusions of earlier studies of population–development interactions were subjected to critical scrutiny. The result was a revisionist rendering that was both surprising and controversial. Specifically, the revisionists downgraded the prominence of population growth as either a major source of, or a constraint on, economic prosperity in the Third World. The basis of this somewhat startling conclusion was the revisionists’ methodology that (a) assessed the consequences of demographic change over longer periods of time and (b) expanded the analysis to take into account indirect feedbacks within economic and political systems. In general, empirical assessments of population growth will be smaller (less negative or less positive) when using the revisionist’s methodology than when focusing on the short run and ignoring feedbacks. On net, most revisionists conclude that many, if not most, Third World countries would benefit from slower population growth.

Julian L. Simon, 1981

No one was more important in stimulating the new round of debates in the 1980s than Julian L. Simon, author of The Ultimate Resource (1981). This book attracted enormous attention, substantially because of two factors. First, it concluded that population growth would likely provide a positive impact on economic development of many developed, and some less developed, countries. Second, the book was accessible, well written, and organised in a ‘debating’, confrontational style. This included goading and prodding, the setting up and knocking down of straw men, and an examination of albeit popular, but somewhat extreme, anti-natalist positions. Simon’s powerful book helped spawn a group of survey articles in the 1980s.

What accounts for Simon’s positive assessments? Simon was an early advocate of evaluating the full effects of population over the intermediate to long run. He argued that the negative ‘direct’ impacts in the short run will probably be moderated, or sometimes overturned, when households, businesses, and/or governments react to changing prices which signal problems of resource scarcity. Two important examples of responses to population pressures can be cited: those relating to technological change and those relating to natural resource scarcity, both highlighted by Simon.

Technological change. Simon hypothesised and attempted to document that the pace of technological change, and its bias, can be stimulated by population pressures. Technological change, in turn, plays a central role in economic growth theory and has been shown in sources-of-growth studies to be a (the?) key to economic growth. Additionally, with respect to population size impacts in general, Simon observes that major social overhead projects (for example, roads, communications and irrigation) have benefited from expanded populations and scale. (For more detail, see section “Technological Change: Density, Size and Endogenous Growth” above.)

Resource depletion. Consider next the impacts of population growth on natural resource depletion. Theoretically an exhaustion of non-renewable resources (for example, coal and minerals) would appear to be inevitable in the long run. However, such a period may be in the indeterminably distant future. By contrast, Simon argued that the most relevant measure of resource scarcity is its price. He prepared many graphs of US non-renewable resource prices (deflated by price indexes in order to focus on ‘real’ resource trends).

Surprisingly, virtually every resource has experienced a declining real price over lengthy periods of time. This means, à la Simon, that resources are becoming more abundant over time. It seems that the more resources are used, the more abundant they become! How can this happen? Simple. A rising resource price, due in part to population pressures, triggers several reactions that reduce or even eliminate the apparent resource scarcity. Specifically, in the short run, rising prices encourage an economising of the resource at every level of production and consumption. In the longer run, rising prices stimulate exploration, new methods of extraction and process, and the search for substitutes.

Nevertheless, Simon recognised that market failures, institutional failures, and political factors can all result in less-than-complete adjustments when population and economic development press against resource availabilities. This is particularly the case with renewable resources (such as rain forests, fisheries, the environment, and so forth) where market or institutional failures are pervasive. Without mechanisms to assign and maintain property rights, internalise externalities, and address free rider problems of public and quasi-public goods, government regulation may be required to safeguard renewable resources over time.

National Academy of Sciences, 1986

Some 15 years after the 1971 National Academy Report that highlighted 25 negative consequences of population growth, a new National Academy Report was released. In contrast to the previous study, the new report was balanced, eclectic and non-alarmist. A careful examination of its bottom line is instructive.

On balance, we reach the qualitative conclusion that slower population growth would be beneficial to economic development of most developing countries. (1986, p. 90; emphasis added)

This qualified assessment reveals key features found in most population assessments in the 1980s. Specifically: (a) there are both positive and negative impacts of demographic change (thus ‘on balance’); (b) the magnitude of the net impacts cannot be determined given current evidence (thus ‘qualitative’); (c) only the direction of the impact from high to low growth rates can be ascertained (thus ‘slower’ rather than ‘slow’); and (d) the net impact varies from country to country. In most cases it will be negative; in some positive; and in others of little impact (thus, ‘most developing countries’).

What accounts for the dramatic turnaround in the two National Academy assessments? Several factors can be advanced. First, the 1986 report extends the short-run time horizon of the 1971 report to examine individual and institutional responses to the initial impacts of population change: conservation in response to scarcity, substitution of abundant for scarce factors of production, innovation and adoption of technologies to exploit profitable opportunities, and the like. These responses are considered to be pervasive and they are judged to be important. According to the report writers: ‘the key [is the] mediating role that human behaviour and human institutions play in the relation between population growth and economic processes’ (1986, p. 4).

Second, the 1986 study was assembled almost entirely by economists whose understanding of and faith in markets to induce responses that modify initial direct impacts of population change is far greater than that of other social and biological scientists.

Third, research accumulating over the 15 years between the two reports revealed a need to downgrade: (1) the concern about non-renewable resource exhaustion; (2) the adverse impact of children on the capacity to save, and in turn to undertake productive investments; and (3) the inability to invest in schooling and health facilities.

Finally, the 1986 Report upgrades the concern about population impacts on renewable natural resources (such as fishing areas and rain forests) where property rights are difficult to assign and maintain. Overuse can result. It is recognised that the problems of overuse are not solely due to population growth per se, but rather institutional failure. Cutting population growth by one half, or even to zero, would not solve the problem. Rather it would slow the process and postpone the date of resource exhaustion. Government policies are needed to account for negative externalities and market failure. Slowing population growth provides time for institutional response.

New Paradigms for Modelling Demography’s Role in Economic Growth: 1990 and Beyond

As noted previously, Kuznets’s empirical finding of an absence of notable negative correlations between population growth and per capita output growth influenced the population debate throughout the 1970s and 1980s. Simple correlations stimulated research during the 1990s as well. This time, however, statistically significant negative correlations during the 1980s drove the discussion. Interestingly, economic–demographic modelling continued in the ‘revisionist’ vein, incorporating positive and negative as well as short- and long-run influences into an economic growth model. The modelling challenge remains one of accommodating correlations that can be negative, positive or insignificant depending upon time and place.

Convergence Growth Models: A Framework for Assessing Demography’s Impact

Renewed interest in modelling the impacts of demographic change on economic growth coincided with the emergence in the economic growth literature of the ‘technology gap’ or ‘convergence’ model. This model, formulated initially by Barro and Sala-i-Martin (1991), has been used widely to explore many hypothesised influences on economic growth, including openness to trade, form of government, and the rule of law. Since this type of modelling highlights the dynamics of the adjustment process, it is particularly relevant to examining the impacts of major shifts in the population’s age distribution associated with birth and death rates that change systematically over the demographic transition. As a result, economic demographers have employed convergence paradigms to explore demographic–economic interactions.

Briefly stated, convergence models focus on the pace at which countries move from their current level of labour productivity to their long-run or steady-state level of labour productivity. The model assumes that all countries converge at the same rate from their current to their long-run levels (which can vary across countries and over time). The greater the productivity gap, the greater are the gaps of physical capital, human capital and technical efficiency from their long-run levels. Large gaps allow for ‘catching up’ through (physical and human) capital accumulation, and technology creation and diffusion across countries and over time. Indeed, many empirical studies indicate that growth rates do slow down as a country approaches its long-run productivity level, especially those studies that provide for country- and period-specific conditions that influence the long-run level of labour productivity.

Since long-run labour productivity is unobservable, empirical implementations of the model substitute a vector of ‘conditioning’ variables thought to influence long-run labour productivity. The actual specification of these conditioning variables varies notably. Consider two of their many representations. The first, by Barro (1997), highlights inflation, government consumption ratios, the rule of law, the form of the political system, terms of trade, human capital, the total fertility rate, and life expectancy at birth (a proxy for health). The second formulation, by Bloom and Williamson (1998), highlights two categories of growth-rate determinants: economic structure variables (natural resources, schooling, access to ports, location in the tropics, whether landlocked, and extent of coastline); and economic and political policies (openness to trade, quality of institutions, and government savings share of GDP). Clearly there are many defensible perspectives on variable choice, and much is yet to be learned about the appropriate configuration of conditioning variables that influence long-run productivity levels.

Alternative Demographic Renderings Within a Convergence Framework

The 1990s witnessed attempts by various researchers to model demography in a manner that accommodates both the insignificant correlations of the 1960s and 1970s as well as the significant negative correlations of the 1980s and 1990s. Three different approaches are described here. All three employ a convergence-type growth model and all employ a broad set of countries spanning the income spectrum.

Modelling through aggregate measures of fertility and mortality. Barro (1997) includes two demographic aggregate measures among his list of conditioning variables, the total fertility rate (TFR) and life expectancy. Barro’s formulation thus has demography impacting the long-run equilibrium level of per capita income. The TFR captures, for example, the adverse capital-shallowing impact of more rapid population growth as well as the resource opportunity costs of bringing up children. Furthermore, while Barro treats life expectancy as a human capital proxy for health, demographers consider it to be a demographic variable. Both are statistically significant, with a higher TFR inhibiting, and longer life expectancy enhancing economic growth.

Modelling through population growth components. Kelley and Schmidt (1995) decompose population growth by examining two components (births and deaths) and by modelling their contemporaneous and lagged impacts. This approach allows for disparate impacts of fertility and mortality as well as negative short-run effects (costs of high birth and death rates) and positive long-run effects (favourable impacts of past births on current labour force growth and declining mortality). Consistent with Kuznets’s earlier work, they found an absence of a net demographic impact on economic growth in the 1960s and 1970 – the separate impacts of births and deaths are notable but offsetting. Consistent with empirical work of the early 1990s, they found negative impacts throughout the 1980s. These negative correlations were in part the result of (a) rising short-run costs of high birth rates, (b) declining benefits of mortality reduction, and (c) insufficient labour force entry from past births to offset these increased costs.

Modelling through differential age-structure growth. In a series of papers beginning in the late 1990s, several Harvard economists argued for a demographic rendering that incorporates not only population growth but also labour growth (see, for example, Bloom and Williamson 1998; and Bloom et al. 2000). They note that, while theorists conceptualise the economic growth process in labour productivity terms, empirical growth models are generally specified in per capita terms. This makes no difference when population and labour grow at the same rate, but does when they grow at different rates.

The authors argue that the post-war period was exactly such a time since during that period demographic transitions took place in different countries at different times and at different paces. At various stages of the demographic transition, the population and working ages (used within this framework as a proxy for labour) can grow at very different rates. In a predictable pattern, the population initially grows faster, then slower, and then faster than the working-aged population during the transition from a high-fertility, high-mortality to a low-fertility, low-mortality demographic steady-state equilibrium. (For an historical evolution of economic, sociological, and biological factors during the demographic transition, see R.A. Easterlin 1978.)

Without allowing for differential growth rates of the population and working ages, demographic coefficient estimates (mainly population growth) will be biased. In that case the population–growth coefficient captures net demographic impacts that can be positive, negative, or neutral, depending upon time and place. Bloom and Williamson (1998) demonstrate this point for a broad cross-section of countries over the period 1965–90 in a convergence model that also includes life expectancy as a human capital variable. Consistent with some studies, their simple demographic rendering results in a positive but insignificant coefficient for the population growth rate. When supplemented by the working-age growth rate, however, that coefficient turns negative and the coefficient for the working-age growth rate is positive, both statistically significant.

Effectively, the Harvard economists append an accounting structure to translate labour productivity impacts into per capita terms. The resulting demographic specification is elegant in its simplicity, incorporating only two demographic variables that have unambiguous predicted coefficient values of – 1 (for population rate of growth, Ngr) and +1 (for working-age population rate of growth, WAgr) when used to expose demography’s impact on income growth per capita relative to income growth per working-age population. In that context, demography exerts its primary impact on the pace at which the long-run equilibrium is reached (Bloom and Williamson 1998, p. 419) rather than on the long-run equilibrium level of productivity.

This is an intriguing specification. The interpretation is clear: if labour force growth exceeds population growth, then the rate of per capita income growth is boosted by demography. The Harvard economists label this phenomenon the ‘demographic gift’ that may be reaped for several decades after the onset of fertility decline as new labour force entrants from earlier large birth cohorts outpace fertility. The ‘gift’ was large throughout the 1965–90 period for Japan and other Asian Tigers because of the early and rapid pace of their demographic transition. Of course, the converse of the ‘gift’ began to be felt in the 1990s as new labour force entry from smaller birth cohorts was outpaced by labour force exit of the ageing population. The model predicts productivity outpacing per capita income growth over several decades into the future in these Asian (and other) countries.

Note that the qualitative predictions are based on theoretically determined coefficients on WAgr and Ngr of +1 and −1, respectively. To the extent that estimated coefficients deviate from +1 and −1, WAgr and Ngr play an additional role in the determination of the long-run productivity level. The Harvard studies provide some guidance in this area. In their earlier study, Bloom and Williamson (1998) estimate coefficients that differ significantly from +1 and −1. However, in a later study that further elucidates the accounting, Bloom et al. (2000) find no significant difference from those values. If that is the case, then the model at once makes an important contribution and is somewhat narrower than many in the literature which admit both short-run and long-run impacts of demographic change as a part of the theoretical structure. Yet modelling demography in growth equations tends to be both imprecise and ad hoc. In contrast, the Bloom and Williamson model is relatively clear in interpretation, and it targets the shorter-run impacts that are of primary interest to policymakers.

The Bottom Line

Bloom and Williamson (1998) estimated that as much as one-third of the average per capita income growth rate in East Asian countries over the period 1965–90 is explained by population dynamics. Kelley and Schmidt (2001) evaluated eight distinct demographic renderings within a convergence model using a consistent set of conditioning variables – those described above for Barro’s variant. Among others, these renderings included Barro’s TFR; a ‘naive’ variant predating the 1990s work that simply includes Ngr; a ‘components’ model (contemporaneous and lagged birth rates and the death rate: Kelley and Schmidt 2001); two variants of the Harvard transitions framework; and demographic extensions to several variants.

Kelley and Schmidt (2001) find that on average, across all eight demographic formulations and over their full 86-country sample (covering the full income spectrum), approximately 21 per cent of the combined impacts on change in the per capita income growth rate is accounted for by changes in the demographic variables in the various models. What is striking about this result is that the 21 per cent is fairly stable across all eight demographic renderings, from one that is quite simplistic (Ngr only) to those that incorporate short-, intermediate- and long-term population effects. On the one hand, this should not be terribly surprising because of the interconnectedness of all of the demographic measures. On the other hand, while population matters, it is still important to determine why.

Although there is an emerging consensus that the magnitude of the impacts of population growth have been sizeable (for example, 21 per cent globally and as much as 33 per cent in East Asia), the reasons why this is the case are still both contestable and not well understood. Are the demographic determinants primarily longer-run impacts, or are they mainly shorter-run transitional dynamics that are diminishing? Will the so-called ‘demographic gift’ of these dynamics in the past reveal themselves as a ‘demographic drag’ in the future, deriving from reduced fertility, slow population growth and ageing? Or will a new mechanism reveal itself? For example, (a) will future modelling better expose the components of labour force change (for example, utilisation rates, age- and/or gender-specific participation rates); and (b) will fertility and mortality be endogenously specified to better reveal the dynamics of the demographic transition about which the field of economic demography has much to say? Whatever the outcome, the stage is set for another round of research, pinning down the results of the past with the goal of understanding the future.

See Also


  1. Arrow, K.J. 1962. The economic implications of learning by doing. Review of Economic Studies 29: 155–173.CrossRefGoogle Scholar
  2. Barro, R.J. 1997. Determinants of economic growth: A cross-country empirical study. Cambridge, MA: MIT Press.Google Scholar
  3. Barro, R.J., and X. Sala-i-Martin. 1991. Convergence. Journal of Political Economy 100: 223–251.CrossRefGoogle Scholar
  4. Birdsall, N., A.C. Kelley, and S. Sinding (eds.). 2001. Demography matters: Population change, economic growth and poverty in the developing world. Oxford: Oxford University Press.Google Scholar
  5. Bloom, D.E., and J.G. Williamson. 1998. Demographic transitions and economic miracles in emerging Asia. World Bank Economic Review 12: 419–455.CrossRefGoogle Scholar
  6. Bloom, D.E., D. Canning, and P. Malaney. 2000. Demographic change and economic growth in Asia. Population and Development Review 26: 257–1990.Google Scholar
  7. Boserup, E. 1965. Conditions of agricultural growth. Chicago: Aldine.Google Scholar
  8. Boserup, E. 1981. Population and technological change. Chicago: University of Chicago Press.Google Scholar
  9. Chenery, H., and M. Syrquin. 1975. Patterns of development: 1950–1970. Oxford: Oxford University Press.Google Scholar
  10. Coale, A.J., and E.M. Hoover. 1958. Population growth and economic development in low-income countries. Princeton: Princeton University Press.Google Scholar
  11. Easterlin, R.A. 1978. The economics and sociology of fertility: A synthesis. In Historical studies of changing fertility, ed. C. Tilly. Princeton: Princeton University Press.Google Scholar
  12. Glover, D.R., and J.L. Simon. 1975. The effect of population density on infrastructure: The case of road building. Economic Development and Cultural Change 23: 453–468.CrossRefGoogle Scholar
  13. Higgins, M. 1998. Demography, national savings, and international capital flows. International Economic Review 39: 343–369.CrossRefGoogle Scholar
  14. Kelley, A.C. 1988. Economic consequences of population change in the Third World. Journal of Economic Literature 26: 1685–1728.Google Scholar
  15. Kelley, A.C. 2001. The population debate in historical perspective: Revisionism revisited. In Population matters: Demographic change, economic growth, and poverty in the developing world, ed. N. Birdsall, A. Kelley, and S. Sinding. Oxford: Oxford University Press.Google Scholar
  16. Kelley, A.C., and R.M. Schmidt. 1995. Aggregate population and economic growth correlations: The role of the components of demographic change. Demography 32: 543–555.CrossRefGoogle Scholar
  17. Kelley, A.C., and R.M. Schmidt. 2001. Economic and demographic change: A synthesis of models, findings and perspectives. In Population Matters : Demographic Change, Economic Growth, and Poverty in the Developing World, ed. N. Birdsall, A.C. Kelley, and S. Sinding. Oxford: Oxford University Press.Google Scholar
  18. Kuznets, S. 1960. Population change and aggregate output. In Demographic and economic change in developed countries, National Bureau of Economic Research. Princeton: Princeton University Press.Google Scholar
  19. Lee, R.D., A. Mason, and T. Miller. 2001. Saving, wealth and population. In Population matters: Demographic change, economic growth, and poverty in the developing world, ed. N. Birdsall, A.C. Kelley, and S. Sinding. Oxford: Oxford University Press.Google Scholar
  20. Malthus, T.R. 1798. An essay on the principle of population, , 1970. Harmondsworth: Penguin.Google Scholar
  21. Mason, A. 1987. National saving rates and population growth: A new model and new evidence. In Population growth and economic development: Issues and evidence, ed. D.G. Johnson and R.D. Lee. Madison: University of Wisconsin Press.Google Scholar
  22. National Academy of Sciences. 1971. Rapid population growth: Consequences and policy implications, vol. 1: Summary and recommendations; vol. 2: Research papers. Baltimore: Johns Hopkins University Press.Google Scholar
  23. National Academy of Sciences. 1986. Population growth and economic development: Policy questions. Washington, DC: National Research Council.Google Scholar
  24. Schultz, T.P. 1987. Schooling expenditures and enrollments 1960–1980: The effects on income, prices and population growth. In Population growth and economic development issues and evidence, ed. D. Gale Johnson and R.D. Lee. Madison: University of Wisconsin Press.Google Scholar
  25. Simon, J.L. 1981. The ultimate resource. Princeton: Princeton University Press.Google Scholar
  26. Simon, J.L. 1996. The ultimate resource 2. Princeton: Princeton University Press.Google Scholar
  27. Srinivasan, T.N. 1988. Modeling growth and economic development. Journal of Policy Modeling 10: 7–28.CrossRefGoogle Scholar
  28. United Nations. 1953. The determinants and consequences of population trends. New York: United Nations.Google Scholar
  29. United Nations. 1973. The determinants and consequences of population trends. New York: United Nations.Google Scholar

Copyright information

© The Author(s) 2008

Authors and Affiliations

  • Allen C. Kelley
    • 1
  • Robert M. Schmidt
    • 1
  1. 1.